Pattern and Trend Analytics

KDB.AI Pattern and trend analytics

Launch

Fast-moving markets pressure firms to optimize strategies, mitigate risks, and resolve trading infrastructure issues. KX helps customers analyze data patterns, detect anomalies, and manage risk in real-time, ensuring that they stay ahead of potential challenges and seize new opportunities. In this demo, we will explore real-world sentiment analysis on market data to provide first-mover advantage. Click here to continue.

KDB.AI - Pattern and trend analytics

In this demo, we are using Python to connect to our underlying data table called “tesla_trade_tss” and query the table data into memory. There are a number of different clients we can use to connect to our kdb+ data tables using KDB.AI - including REST, Python and q. Click here to continue.

KDB.AI - Pattern and trend analytics

We're looking at TSLA's live pricing data in real-time, and monitoring for anomalous price movements. We can search across our historical data to understand patterns of anomalies and perform similarity search against real-time price fluctuations for those patterns of interest. We’ll return events most similar to our search pattern, and highlight those K-nearest neighbors as they happen, showing anomalies of interest. Click here to continue.

KDB.AI - Pattern and trend analytics

This chart shows live pricing data, updated every second over a three-minute sliding window. Rules are in place to understand when our pricing curve becomes anomalous, compared in real-time against our nearest-neighbor similarity search results. Because this time series data does not require LLM embedding, we achieve superior real-time performance. Click here to continue.

KDB.AI - Pattern and trend analytics

When an anomaly has been detected, the curve is highlighted in red to indicate an unexpected price movement. By performing similarity search against our real-time data, we can anticipate anomalous market behavior - fault conditions where behavior is understood, but occurrence is unexpected. Click here to continue.

KDB.AI - Pattern and trend analytics

As reflected in our chart, Elon Musk tweeted about taking TESLA private, causing TSLA prices to climb. KDB.AI performs similarity search on real-time curve data against what we know to be true historically. It helps to anticipate extraordinary market events. Click here to continue.

KDB.AI - Pattern and trend analytics

We can also search for smaller micro trends. In this example, we narrow our window of focus to search over one minute's worth of data. KDB.AI does not place limitations on how wide or narrow we define a search pattern of interest. Click here to continue.

KDB.AI - Pattern and trend analytics

At 2:08pm that same day, Nasdaq halted trading of TESLA stock. This is reflected in our real-time data, and flagged by anomaly detection. Click here to continue.

KDB.AI - Pattern and trend analytics

We can also perform simultaneous curve analysis on several series of data. Here we monitor price and volume, conditioning against anomalous behavior. Having this data reflected on a single chart provides context to what we’re seeing. Click here to continue.

KDB.AI - Pattern and trend analytics

We can see how pricing and volume are correlated in the moment following Musk's tweet. Continuous monitoring of market conditions allows us to proactively diagnose and address order performance issues across our data feeds, increase customer confidence, and identify bad actors via algorithm execution, while searching for market abuse and toxic order flow. By detecting these changes as they’re happening, we create a moment to respond and react to market opportunities with competitive advantage. Click here to continue.

KDB.AI - Pattern and trend analytics

We can also use similarity search for pattern classification. In this example, we take an ambiguous market event, and we perform similarity search across the history of our data to determine market sentiment based off of previously occurred events. We want to determine the nature of this 'M-Shaped' curve as reflected in our historical pricing data, so that we can understand how to react when we encounter this signal in our live pricing data. Click here to continue.

KDB.AI - Pattern and trend analytics

KDB.AI allows you to search for multiple patterns at once, helping to identify complex signals simulating real-world market behavior. At terabyte scale for historical data, query speed and performance are paramount. By using the power of kdb+’s massive columnar, in-memory data engine, we present a solution that is uniquely suited to the challenge. Click here to continue.

KDB.AI - Pattern and trend analytics

With KX you can

• Detect patterns with temporal similarity search (TSS) for better trading decisions and reduced risk exposure

• Run simultaneous searches on millions of data points in seconds, improving your time-to-decision

• Reduce time-series data windows by over 99% with advanced compression, enhancing search performance

• Generate signals from patterns in historical data to capitalize on market opportunities

• Identify and correct data feed issues fast, minimizing risk and ensuring operational continuity

• Execute complex queries rapidly with a high-performance engine, improving data query efficiency

KDB.AI - Pattern and trend analytics

Explore other KDB.AI use cases

Learn more about KDB.AI with our free course on the KX learning hub.

Get started with KDB.AI

Connect with developers using KDB.AI on our discussion forums.

Get hands-on experience with KDB.AI by activating our free cloud edition.

Book a 30-minute demo and discuss your analytical needs with one of our sales team.